Predicting Porosity Formation in Laser Powder Bed Fusion using In-situ Data from a Two Wavelength Imaging Pyrometer

The goal of this work is the detection of pore formation in laser powder bed fusion (LPBF) using a in-situ two wavelength imaging pyrometer. To realize this goal, in this work we built a large Inconel 625 part measuring 10 mm X 10 mm X 165 mm (build height) under seven different conditions of laser power and velocity. During the process, in-process Melt Pool temperature and Melt Pool shape measurements were acquired using a ThermaViz two-wavelength pyrometer. The porosity in the part was measured using offline X-ray computed tomography. The porosity in different sections of the part were correlated with the meltpool features (shape, temperature, and spatter characteristics) using a variety of supervised machine learning approaches. The resulting prediction accuracy exceeded 90% (statistical F-score). The components of this work will be applied in real-time during the deposition process to alter the host machine if process conditions are favorable for high or low quality.